fully developed a software system by virtue of Windows
98-based VCþþ. The system integrates the technologies of modern signal analysis and processing and ultrasonic imaging, and realizes the quantitative, qualitative and location analysis of the defects. After the ultrasonic signal is converted to digital signal, a key question is to improve the signal-to-noise (STN). Using frequency and wavelet analysis technologies, the authors structure the filters, realized the signal separation from the noise, and estab- lished a sound foundation for the follow-up process. According to the scanning path and imaging principle, the images are reconstructed using the acquired digital signal. The C, B and P imaging have been realized for complex surface parts. The intelligent defects recogniza- tion technology based on the manual nerve network tech- nique, can automatically classify defects, and give out the evaluation results. In the methods of classification and evaluation: one is to directly recognize and evaluate the obtained digital signal; whilst the other is to recognize and evaluate the image. The former is succinct and speedy, whilst the latter is viewed directly